Note: This is a version of the text which is currently in press at the journal Quaternary International. The paper was originally presented at the Arctic Science Summit week in Prague in April 2017, in a session organized by Peter Jordan and Sean Desjardins.

Colin D Wren, University of Colorado, Colorado Springs

Andre Costopoulos, University of Alberta

Maclean Hawley


We show that known archaeological sites in the Wemindji area of central eastern James Bay in Canada are more likely to show higher shoreline stability within a 2km radius than randomly selected locations. We show that archaeological sites are surrounded by areas of greater diversity in shoreline stability within a 20km radius than random locations. We propose that a cultural algorithm guides the selection of occupation sites in the region, including by today’s Cree occupants, which identifies islands of stability in a sea of change in order to minimize the need for residential mobility while maximizing the diversity of resources available through logistical trips. This algorithm favours residential stability and logistical mobility, and evolved in the context of rapid shoreline displacement caused by post-glacial isostatic land-uplift.

Funding: This project was funded by the Social Science and Humanities Research Council of Canada under the Insight Development program.


The territory of the Cree Nation of Wemindji is located about 53 degrees N along the central eastern shore of James Bay in Northern Quebec, Canada (Fig. 1). It has seen intense land uplift resulting in an average vertical displacement of 2.2 m/century since deglaciation (Pendea et al 2010). Due to the relatively flat landscape, this vertical displacement rate created a horizontal westward displacement of the James Bay shoreline by up to 5 km/century (Wren et al 2014). This displacement is still a feature of the life of the Wemindji Cree today and relates to a number of their environmental adaptations (Sayles 2015). As the Wemindji Cree put it, “the land grows”.


Figure 1. Map of eastern James Bay today including coastal Cree communities and Old Factory Lakewhich is mentioned in the text. The estimated position of the Sakami Moraine marks the approximate maximum eastern shoreline of James Bay during its Tyrell Sea phase.

The earliest occupations of eastern James Bay date to 4200 BP (Izaguirre and Denton 2015). Despite significant archaeological work related to hydro-electric projects since the 1960s, relatively little has been published on the region and it remains underexplored.

Due to poor organic preservation in the Boreal forest, archaeological remains are usually limited to small concentrations of lithic debitage and tools, ash and/or charcoal indicating hearths, and occasionally more substantial stone hearths (e.g. FeGj-2, Wren et al 2014). Until recently, archaeologists interpreted the pre-European contact period of the region as one of cultural continuity and stasis (Bracewell 2015), characterized by low density, highly mobile, and interior adapted hunter-gatherers (e.g. Wright 1972). Recent work has demonstrated greater variability in the region’s past, including variations over time in density, mobility, long-distance trade, and distribution of the populations (Holly and McCaffrey 2012, Izaguirre and Denton 2015, McCaffrey 2011, Wren et al 2014, Wren et al 2012).

Settlement choice under changing environmental conditions

Recent archaeological literature on settlement choice suggests that humans respond to environmental change and uncertainty by adopting risk reducing settlement patterns (Morgan 2009, Burke et al 2017), and that reliance on logistical, rather than residential mobility, reduces overall risk for hunter-gatherers (Grove 2010). While environmental change does not determine human settlement patterns under conditions of change and uncertainty (Crema 2013), it is clearly an important constraint on settlement choice. Here, we test whether the placement of archaeological sites on the landscape of Wemindji territory conforms to a logistical mobility based risk reduction strategy in the context of environmental change. Although, logistical strategies are common in higher latitudes with marked seasonality in general, Binford (2001) reports a variety of logistic and residential mobility strategies for known groups in northern Quebec.

Because the landscape is undergoing rapid shoreline displacement, reduction of residential mobility should favour the selection of sites that remain coastal for long periods of time. The same shoreline displacement generates an ecological succession across the landscape that results in change in available resources over time. Reliance on logistical mobility should favour the selection of sites surrounded by varied resources, and in the Wemindji context, for sites surrounded by variable shoreline displacement and its accompanying ecological succession. Diversity of shoreline displacement rates within a small radius generates a diversity of environments because different locations will be at different points of the standard ecological successions from coast, to salt-marsh, to fen, and finally to bog or forest, depending on local topography (Pendea 2011). In a logistic strategy, access to a diversity of habitats within a day’s travel would ensure access to a wide variety of the resources needed in the relatively sparse Boreal environment without the high cost of residential travel (Binford 1990).

We therefore hypothesize that archaeological sites in Wemindji territory will be coastal for relatively long periods of time, while being surrounded by areas in which ecological change is fairly rapid. In other words, we expect archaeological sites to be islands of coastal stability surrounded by diverse and changing catchments (Fig. 2).


Figure 2. Schematic diagram of our hypothesized model. Hypothetical coastal site is selected in a location with a long duration of coastal access but surrounded by a more variable landscape where a diverse set of resources are available.


We compared known archaeological material concentrations on the landscape, hereafter referred to as sites, with randomly selected points and compare them for coastal duration and variability of coastal duration within radii of 2, 5, 10, and 20 kilometers. Randomly selected points act as our proxy for locations not selected for occupation, i.e. non-sites, though since these were not ground-truthed for absence of cultural material (and the majority are not easily accessible either), we refer to them as pseudo-non-sites.


We collated all known archaeological sites within the study region classified as “amérindien préhistorique” within the Inventaire des sites archéologiques du Québec including sites from our own surveys (Wren et al 2014, Wren et al 2012). The study region extended from the James Bay coast in the west to the Sakami Moraine to the east. We focused on the central James Bay coast where our surveys had taken place roughly, 40 km to the north and south of the Wemindji Access Road (Fig. 1). We downloaded digital elevation model data (DEM) from GeoGratis at an approximate cell resolution of 25m. Using the elevation and date of emergence in cal yr bp from a prior study of shoreline uplift (Pendea et al 2010), we created a simple biplot and fit a second order polynominal equation to these data. The equation thus allows us to accurately estimate the date of coastal emergence for any given elevation (Eq. 1). We applied this equation to the DEM using a GIS raster calculator, where the elevation raster is used in place of , to derive a raster of date of coastal emergence in cal yrs bp (Fig. 3).

y = -0.1425x^2 + 65.653x

where y is the emergence date in years cal bp and x is elevation, or DEM raster, in m asl.

Using a focal statistics tool, we calculated a new raster representing the per cell minimum emergence date within each of the radii mentioned above, and subtracted this from the emergence date. This difference gave us a raster of the duration in years for which any given location remained within that radius of the retreating James Bay coastline (Fig. 4). We used the focal statistics tool again to calculate the standard deviation of coastline duration using the same radii (Fig. 5). Higher standard deviations indicate more variability in coastline duration and therefore more ecological variability due to the standard ecological succession. We then extracted the local raster value for each of these variables using the known site locations and pseudo-non-site points.


Figure 4. The duration of time (in years) each part of the landscape remains within a 5km radius of the James Bay coastline as it retreats westwards over time. The darkest red areas remained within a 5 km distance of the coastline for over 5000 years, whereas the darkest blue areas would have seen the coastline retreat out of accessible range within a person’s lifetime. Black dots are known prehistoric sites.


Figure 5. Variability of shoreline proximity duration calculated by taking the standard deviation of the duration shown in Figure 3 within a 5km radius. Darker red areas represent areas with a wider variety of shoreline displacement rates, and therefore a wider variety of ecological succession patterns within an accessible range. Grey dots are the randomly generated pseudo-non-site locations used in the analysis.


Known concentrations of archaeological material, that is sites, are within 2km of the coastline for 66% longer on average than pseudo-non-sites (961 years vs 579 years, Fig. 6). This drops to 12% more within a 5km radius (1134 years vs 1010 years). Within a 10 km radius, sites and pseudo-non-sites are coastal for about the same duration (1452 years vs 1457) and within a 20km radius, coastal duration is 25% lower for sites than for pseudo-non-sites (1953 years vs 2347 years).

Duration variability is 77% higher for sites than for pseudo-non-sites (AvgStDev = 536 vs AvgStDev = 304, Fig. 7) within a 2km radius, 50% higher for sites for a 5km radius (AvgStDev = 651 vs AvgStDev = 434), 31% for a 10km radius (AvgStDev = 764 vs AvgStDev = 583), and 21% higher within a 20km radius (AvgStDev = 982 vs AvgStDev = 811). Sites are coastal for markedly longer and have markedly more diverse environments than pseudo-non-sites within a 2km radius. Both values are higher for sites than for pseudo-non-site locations for all radii considered.


Figure 6: The lines on this bumps chart show the degree of change from non-sites to sites for each of the tested radii. Note that the increase in coastal duration from non-sites to sites is strongest with a 2km radius and decreases with greater radius.


Figure 7: Variability in coastal duration, measured as standard deviation within the specified radius, increases strongly for all measured radii.


Figure 8: Amount of difference between the average of pseudo-non-sites and three categories of known points: archaeological sites, the Hudson’s Bay trading post on Old Factory Island, and modern day Wemindji. Positive direction on each axis shows known sites have the greater value.


These results are consistent with a culturally evolved set of settlement location criteria and decision-making processes, ie. a selection algorithm, that values direct access to the coast, i.e. within a day’s round-trip travel, and variability within a logistical catchment. When compared to random locations, the preference for coastal stability for sites is strong within a 2km radius but falls off rapidly. It is still slightly higher for sites if we look at a 5km radius, but seems irrelevant at the 10km and 20km radii. This makes sense if the goal of selecting a coastally stable site is for immediate access to the coast. For immediate access, the presence of coast within 2km would be much more significant than its presence within 20km.

Immediate access would be important for accessing costal food resources. In the past, as today, James Bay supports migratory birds, fish, shellfish, sea mammals, marine flora, and other wildlife. Although soil conditions rarely permit preservation of organic materials in the region, Izaguirre and Denton (2015) confirmed early coastal use ca. 4000 cal yr bp at a recently excavated site in southern James Bay.

Immediate coastal access would also enable north-south canoe travel along James Bay’s coast. As most rivers run east to west, north-south travel along James Bay enabled access to other watersheds. Long-distance travel north-south along James Bay is supported by the presence of Nastapoka chert in several sites in our sample, which would have been extremely difficult to transport through the interior ca. 400 km from its source on Hudson’s Bay(Wren et al. 2012; 2014).

As we expand the radius around a site, coastal stability becomes less important, but variability of coastal stability within the radius retains its importance. Variability in shoreline displacement rate is 77% higher for sites than for pseudo-non-sites at a 2km radius. This falls off to 21% at a 20km radius, but remains higher for sites than for pseudo-non-sites for all radii considered. Our proposed cultural algorithm favours areas where the shoreline is more stable (i.e. will remain within 2km of the coast for about 1000 years as compared to pseudo-non-sites’ average of 600 years), and where there is a diversity of shoreline displacement rates within a 20km radius to provide access to more diverse resources.

An alternative hypothesis to our cultural site selection algorithm is that coastal stability simply promotes the archaeological detection of remains of human activity. If people were accessing the James Bay coast without consideration of coastal stability then material evidence of human use would form a more substantial presence on the most stable sections of coastline through more frequent re-use. This would make sites more likely to be detected with higher coastal stability but should not affect the variability of coastal stability, especially not so strongly and consistently through the wide range of radii we sampled. We therefore find the selection algorithm hypothesis to be more likely.

Applying the algorithm to the relocation of Wemindji

Starting in the 18th Century, the Hudson Bay Company (HBC) traded furs at the mouth of the Old Factory River, south of the present village of Wemindji. By the 1930s, it had established a full time trading post and a small Cree community had grown around it. In the mid-1950s, a community elder had a dream that the community should relocate to the present site of Wemindji (Cree Nation of Wemindji 2010). Figure 8 shows clearly that the new site of Wemindji is much more like known archaeological sites than the site chosen by HBC for its trading post. Wemindji has much higher coastal durability and variability in shoreline displacement at all radii than Old Factory, and even than the known archaeological sites. HBC’s trading post has lower coastal durability and lower environmental diversity than the random locations at all radii tested.

Old Factory River, being a major trade route and connected to the resource rich Old Factory Lake, made sense for the fur trade (Wren et al 2014), but did not fit traditional Wemindji site selection criteria. Our analysis suggests that the elder’s selection of the new site of Wemindji was consistent with an evolved cultural algorithm that favours immediate access to the coast for long durations, as well as ecological variability within a catchment of at least 20km (Fig. 8). After all, the elder presumably would not have dreamed about a location she was not already intimately familiar with. While this is simply an anecdote at this point, it certainly fails to invalidate our main thesis.

At a broader level, the fact that settlement choice in the region might be strongly influenced by the distribution of shoreline displacement rates at any one point in time, helps to explain that settlement mobility and density varied through time in ways that were not anticipated by the traditional model of the Shield Archaic (Wright 1972).


We show that known occupation sites in the Wemindji area of eastern James Bay, including the modern village of Wemindji, fit certain settlement choice criteria that emphasize extended direct access to the coast and diversity of shoreline displacement rates within radii between 2km and 20km. We propose that this pattern of settlement selection reflects the long-term evolution of a cultural algorithm that favours residential stability and logistical mobility in order to reduce risk in the context of rapid environmental change.

This model has the advantage of being testable using datasets from other regions undergoing post-glacial shoreline displacement and can eventually be extended to other dimensions of environmental stability, such as rainfall.


Binford, L.R., 1990. Mobility, Housing, and Environment: A Comparative Study. Journal of Anthropological Research 46, 119–152.

Binford, L.R., 2001. Constructing frames of reference: an analytical method for archaeological theory building using hunter-gatherer and environmental data sets. University of California Press, Berkeley.

Bracewell J 2015. The Infertile Crescent Revisited: A Case (Study) for the History of Archaeology. Bulletin of the History of Archaeology. 25(2), p.Art. 3. DOI:

Burke A, M Kageyama, G Latombe, M Fasel, M Vrac, G Ramstein, PMA James 2017. Risky business: The impact of climate and climate variability on human population dynamics in Western Europe during the Last Glacial Maximum, Quaternary Science Reviews 164:217-229.

Crema E 2013. Cycles of change in Jomon settlement: a case study from eastern Tokyo Bay, Antiquity 87:1169-1181.

Grove M 2010. Logistical mobility reduces subsistence risk in hunting economies, Journal of Archaeological Science 37:1913-1921.

Holly DH and M McCaffrey 2012. Rethinking Eastern Subarctic History, in TR Pauketat (ed) The Oxford Handbook of North American Archaeology, Oxford University Press :124-134.

Izaguirre, D., Denton, D., 2015. Résultats de l’intervention archéologique sur le site EhGo-l, Waskaganish, Québec: Rapport d’etape, saison 2014. Gouvernement de la Nation crie et la Nation crie de Waskaganish, Waskaganish, QC.

McCaffrey M 2011. Ancient Social Landscapes in the Eastern Subarctic, in KE Sassaman and DH Holly (eds) Hunter-Gatherer Archaeology as Historical Process, Tuscon: The University of Arizona Press :143-166.

Morgan C 2009. Climate change, uncertainty and prehistoric hunter-gatherer mobility, Journal of Anthropological Archaeology 28:382-396.

Pendea I.F., A Costopoulos, C Nielsen (CD Wren), GL Chmura 2010. A new shoreline displacement model for the last 7 ka from eastern James Bay, Canada. Quaternary Research 73 :474-484.

Pendea, I.F., 2011. Mid to Late Holocene Coastal Landscape Change in Eastern James Bay. PhD Dissertation. McGill University, Montreal.

Sayles JS 2015. No Wilderness to plunder: Process thinking reveals Cree land-use via the goose-scape, The Canadian Geographer 59:297-303.

Cree Nation of Wemindji. 2010. Wemindji Turns 50: A Community Where Tradition Lives On. Farrington Media: Milton, ON.

Wren CD, A Costopoulos, IF Pendea, I Paberzyte, J  Bracewell, S Vaneeckhout, J Okkonen, CJH Ames, E Hulse, 2014. Archaeological and paleo-environmental survey in Wemindji, James Bay’s paleo-coastal region (2005-2010). Unpublished Report. Ministère de la Culture et des Communications, Gouvernement du Quebec. Montreal, QC.

Wren CD, M Hawley, M King, A Costopoulos, IF Pendea, 2012. Archaeological and paleo-environmental survey in Wemindji, James Bay’s paleo-coastal region. Unpublished Report. Ministère de la Culture et des Communications, Gouvernement du Quebec. Montreal, QC.

Wright JV 1972. The Shield Archaic, Ottawa: National Museums of Canada.

2 thoughts on “Preprint: Settlement choice under conditions of rapid shoreline displacement in Wemindji Cree Territory, subarctic Quebec

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s